Embedded Processor Watch

Senior Editor: Tom Halfhill
Contributors to this issue: Peter N. Glaskowsky, Senior Editor; Michael Slater, Principal Analyst

In This Issue:

• Silicon Magic: DVine-ly Inspired?
• Embedded Internet: What Problem Are We Solving
• Microprocessor Forum Call for Proposals
• WinHEC's Not All That's Hot in N'Awlins

Silicon Magic: DVine-ly Inspired?

By Peter N. Glaskowsky

There's a new entrant in the already crowded field of chip multiprocessor (CMP) media processors. Silicon Magic's DVine architecture, named after its combination of embedded DRAM and vector engines, is simpler and narrower in focus than most media processors, which are often meant to cover a wide range of potential applications. DVine has a single purpose: to replace fixed-function audio and video codecs in consumer electronics. Silicon Magic says its approach will increase performance and flexibility while cutting development time and chip count.

DVine combines compute modules (CMs), each of which consists of a simple 32-bit RISC processor and a 16-way vector-processing engine, with embedded-DRAM arrays managed by programmable memory controllers capable of simple operations such as interleaving, interpolation, and rounding. DVine chips also include an external bus interface, multiple 128-bit internal data buses, and a data-flow controller. Because of Silicon Magic's more limited focus, DVine lacks integer multiplier units and has no support for floating-point calculations. Most other media processors offer both of these capabilities.

For some purposes, the DVine architecture may provide the perfect combination of features. Silicon Magic says a DVD decoder may be implemented with just two CMs and 2M of embedded DRAM in a chip that should cost only a few dollars to manufacture in high volume. Adding incremental features to such a design can be as simple as designing a new chip with more CMs and more memory and writing the necessary code. There are many other ways to build DVD players, however, and some are no more expensive than the DVine approach. Even if the development cost of the ASIC solution is slightly higher, the lower manufacturing cost of an ASIC will save money in the long run for products manufactured in high volume. Silicon Magic's DVine may need divine guidance to succeed in a market already full of media processors. (The full version of this article is available online to Microprocessor Report subscribers at http://www.MDRonline.com/mpr/h/2000/0327/141302.html).

Embedded Internet: What Problem Are We Solving?

Editorial By Michael Slater

There is a lot of talk these days about embedding Internet connections in all kinds of appliances. Not just Internet appliances, but ordinary appliances -- like dishwashers and coffee makers. No doubt there will be some compelling applications for such things, but I'm concerned that there is more technology push going on here than consumer pull. Are any of these products going to actually enhance our lives? I'm skeptical.

The Internet makes it possible to connect virtually every electronic device on the planet that has even a modest amount of computing power. Software developers are putting TCP/IP stacks into 8-bit microcontrollers, and others -- such as iReady -- are putting Internet interfaces directly into logic gates (see Embedded Processor Watch #22, http://www.MDRonline.com/epw/issues/epw_22.html). Web servers can be built on tiny PC boards. The magic of semiconductor technology is driving the cost of an Internet connection to just a few dollars.

For industrial automation, pervasive Internet connections are compelling. For office equipment and building automation, they probably make sense. But I'm having trouble with the kinds of applications that advocates are discussing for consumers.

At one recent conference, two panelists discussed -- with straight faces -- how your alarm clock could communicate with your coffee maker so the coffee would get started at the right time. Am I missing something here? I personally don't have a lot of trouble turning on the coffee maker manually. Having to wait three minutes for the coffee to brew is not a problem worth solving.

Next the discussion turned to how your washing machine could be automatically updated by the manufacturer -- perhaps to add a new cycle to handle a new type of fabric. I have a hard time believing there is much value in this. And it seems all too likely the manufacturer would use such "upgrades" to try to extract ongoing revenue from its customers.

Then there is the much-discussed smart refrigerator that notices when I've run out of milk and orders more. But what if I didn't want more milk? Maybe I'm going out of town for a week. (And please don't tell me that it would know that from my calendar and respond accordingly.) On the other hand, a kitchen appliance that will -- on request, not automatically -- scan the bar codes from things I'm about to throw away and add them to my shopping list could be genuinely useful.

In the discussions about intelligent appliances, there seems to be remarkably little appreciation for real consumer benefits -- or for the tradeoffs between simplicity and convenience. Having my alarm clock communicate with my coffee maker is far more likely to create frustration than it is likely to improve my life. The more communication and automation that is attempted, the more potential there is for the system to automatically do something I don't want it to do. And there is no existing proof that the user interface can be crafted in such a way that ordinary people will find it easy to override the defaults and correct anomalous automatic behavior.

I remain a big believer in information appliances -- devices that allow us to interact easily with digital media. I think such devices can offer real, compelling consumer benefits: I would like to be able to read news from the Web at the dining table; to look up phone numbers electronically instead of in fat, outdated phone books; and to find nearby restaurants (and reviews of them) when I'm traveling. There is real benefit in connecting my TV to the Internet so it knows what shows are on when. The CD changer too can benefit from a Web connection that would enable it to know the name of the CD, the artists, and the titles. There may be a few isolated instances where some benefit can be had by connecting ordinary devices to the Internet -- but I'm happy with my dumb appliances. Designers should focus on solving real problems, not on pushing Internet technology into every conceivable corner. (Michael Slater welcomes your comments on his editorial: mailto:mslater@mdr.cahners.com).

Microprocessor Forum Call for Proposals

Cahners MicroDesign Resources is accepting presentation proposals for its thirteenth annual Microprocessor Forum, October 9-13, 2000, at the Fairmont Hotel in San Jose, California.

We will consider proposals from companies making announcements of new high-performance microprocessor technology. Presentations must include significant technical detail and include new information not previously disclosed. Microprocessors disclosed in detail for the first time will be given preference. Areas of interest include processors for PCs, processors for workstations and servers, embedded processors, DSPs, and programmable multimedia processors.

Your abstract must be submitted electronically to: mailto:mpfprogram@mdronline.com

Deadline for proposals is June 1, 2000.

All presentation proposals will be kept confidential. Proposals may be preceded or accompanied by appropriate non-disclosure agreements. For more information, go to http://www.mdronline.com/mpf/call.html.

WinHEC's Not All That's Hot in N'Awlins

Don't let the delicious gumbo, etouffee, and jambalaya in the nearby French Quarter distract you from the spicy technology being served up at WinHEC 2000 in New Orleans, April 25-27. WinHEC 2000 will have twice as many technical tracks as last year's show, punctuated by keynotes from the ubiquitous Bill Gates, Intel VP Pat Gelsinger, and MDR's own Michael Slater -- whose presentation is "Selecting Microprocessors, System Logic, Memory, and I/O Approaches for Windows PCs." For more information, visit the WinHEC Web site at http://www.microsoft.com/winhec.